DUNE
From GridPP Wiki
Revision as of 08:03, 20 April 2018 by Andrew Mcnab 40269d547a (Talk | contribs)
Contents
Overview
The DUNE experiment and DUNE Wiki.
DUNE in the UK
GridPP point of contact: Andrew McNab
Mailing list: dune-uk-comp@cern.ch (join in the usual way or ask Andrew)
Supporting DUNE via GridPP dirac
This route was used Jan-Apr 2017.
- To enable DUNE at your site, please enable it according to DUNE config.
- Please don't forget to enable the pilot role.
- If you want to test your site, please email Simon and Daniela at Imperial.
- At the moment DUNE does not request storage at the sites they run. They currently use storage at Sheffield with a minor outpost at Imperial
- If the accounting data are to be believed, we have so far run one million DUNE jobs in the UK :-) (May 2017)
Supporting DUNE via the Fermilab ITB
This route is still (April 2018) being used, almost exclusively at Imperial and Sheffield. We still (April 2018) need to check the details and for updated requirements/desirables from the FNAL side.
- DUNE might contact your site, or failing that, please contact Ken Herner at Fermilab.
- DUNE will usually request 8 core jobs. At Imperial we are running DUNE as single core jobs, as they are currently mainly running single core jobs within their 8 core slots and this model has proven very inefficient when used by CMS. Which configuration is run is obviously a site decision.
- Some DUNE users will use proxies issued by cilogon-basic, this is not a proper CA (details in the ticket below if you are interested). For all practical purposes this means glexec will not work and should be turned off on the factory side.
- Here is the Imperial OSG ticket, where most of these issues are discussed.